Potential updates for `electromagnetics.frequency_domain`

[sgkang]

These are thoughts that I had while integrating emg3d into SimPEG

• [ ] General direction for receiver objects (currently it only supports x, y, z)
• [ ] Line source implementation (I worked on this with @micmitch and @jcapriot, so we should bring that in) / the same implementation can be used for receivers.
• [ ] Handling of complex data (currently, we assume the data are real-valued); this could save some time.
• [ ] When addressing an adjoint operation, we use transpose, but this is needed to be Hermition (it works, but might be better to make change...)